In 2025, corporate energy management is more complicated than ever.
First, there’s vendor sprawl: different solar or demand response providers by region, and procurement contracts scattered across multiple brokers. Then there’s increased volatility: rising demand, worsening grid constraints, and spiking capacity prices.
Udit Garg, the VP of product at Arcadia, kept hearing a similar request from corporate energy managers: “I’m doing this on hard mode — can you do this on easy mode for us?” Today, Arcadia is launching an “easy mode” offering that includes a new suite of AI-enhanced tools for corporate customers navigating the increasingly complex electricity market.
It’s a significant expansion for the company, which initially built products to connect residential consumers with clean electricity, and then later became a back-end data provider to the energy industry. And it comes about nine months after the company netted $50 million in growth financing from new and old investors, and closed a new $30 million credit facility with J.P. Morgan and TriplePoint Capital.
In a piece published in Latitude Media this week, Arcadia CEO Kiran Bhatraju argued that the combination of supply scarcity and market volatility “will create significant headaches and complexity for energy buyers across the country.”
In an interview, Garg emphasized the impact of that complexity on Arcadia’s customers. “We’re about to enter a world of energy scarcity,” he said. “What used to work — refreshing your strategy every couple years, using spreadsheets and people — that’s not going to cut it anymore.”
He pointed to recent shocks like a tenfold spike in capacity prices in PJM as evidence of the new reality: “You need to be in the game with locked up contracts, otherwise the data centers might suck it all up and you’re left being a price taker.”
For years, Arcadia operated as what Garg calls “the Plaid for energy,” providing back-end data services to solar companies, energy management firms, and electric vehicle companies. The company now processes about 3 million bills a month across 5,000 utilities — creating a massive data set that Arcadia is now applying to energy management.
The new enterprise platform combines three core offerings: utility bill management, energy procurement, and sustainability reporting. The system automatically handles bill payment and auditing, evaluates onsite and offsite procurement options, and feeds standardized energy data into sustainability reporting platforms.
As Bhatraju noted, Arcadia’s Signal database tracks every rate in the U.S., and over the past five years has documented a 400% increase in commercial time-of-use rates offered by utilities. By arbitraging rates, companies can save hundreds of thousands — perhaps millions — of dollars per year.
According to Garg, this unified approach allows customers to move beyond the traditional point-solution model where different vendors handle different aspects of energy management.
When analyzing a large bank’s portfolio, for instance, Arcadia found that solar developers were proposing to maximize installations across all available rooftops and parking lots. Running the data through its optimization engine, the company recommended a different strategy: outfit half the rooftop with solar panels, add battery storage, supplement with community solar credits, and fill the remaining needs with clean power procurement contracts.
“A solar company would have told them to install carports, install just max all the solar you can on all your land,” Garg said, but this mixed approach delivered roughly double the value of the all-solar strategy, he claimed.
Finding ‘pressure points’ where AI is valuable
The platform’s AI capabilities build on Garg’s experience with predictive maintenance systems for the smart grid at the enterprise software company C3 AI. Rather than broadly applying generative AI, Arcadia has identified specific “pressure points” where machine learning can improve outcomes.
The company employs AI across three distinct layers of its stack. At the data acquisition level, Arcadia already uses large language models in a secure way to interpret bills.
“We’re getting bills and using LLMs to help us generate our code because it’s much more cost-efficient,” Garg explained. “The value is actually in compressing the work to generate the bot code and then running it millions of times. It’s much cheaper to have our own software do it than paying OpenAI or Anthropic to do it.”
In the middle layer, AI acts as a quality control system for data cleaning and normalization.
“We can’t go to a customer and say ‘an AI model got it 95% correct, and we don’t know which 5% is wrong,'” Garg noted. Instead, the company has built hyper-targeted models to catch common data errors, gap-fill missing information, and join disparate datasets. “When this tariff database says this thing, we run it through an LLM to see if they mean the same thing. Then we can join those data sets together.”
The technical foundation has evolved significantly from earlier approaches. During his C3 days, “we had data scientists coding things up on the side and running things and saving those outputs into a database,” Garg explained. “But now Snowflake, AWS, Databricks — there’s so many companies and tools you can use to write your model and they just naturally scale and operationalize those for you.”
To support the expansion, Arcadia continues to build a data science team with energy industry experience, while encouraging all of its engineering teams to experiment with AI integration. “I want many of our engineers experimenting with this and becoming more fluent with it,” Garg said, “because my vision is they’re all thinking about how they can embed AI into every little thing they’re doing.”
As a result of both market demands and the growing complexity of corporate energy transitions, Garg expects this enterprise expansion could lead to “a doubling in our growth over the next two to three years.”
